Abstract

e16571 Background: Erdafitinib (erda) is a fibroblast growth factor receptor (FGFR) inhibitor used in previously-treated FGFR2/3-altered locally advanced/metastatic urothelial cancer (LA/mUC). FGFR2/3 alterations (alts) are present in 15-20% of patients (pts) with LA/mUC. We aimed to investigate “real-world” outcomes of erda in LA/mUC. We hypothesized that efficacy of erda would be similar to clinical trial data. Methods: Pts with FGFR2/3 alts who received erda were included. Overall survival (OS) and progression-free survival (PFS) measured from erda start were assessed with KM method and group comparison by long rank test. Categorical variables and observed response rate (ORR) were compared using Fisher’s exact test with Baptista-Pike method for 95% confidence intervals. Cox regression model used to describe PFS and OS for subgroups. Results: Among 633 pts from 16 US sites, 487 (77%) had NGS data, 94 (19%) had FGFR2/3 alts and 39 (41%) received erda. In erda treated pts, median age was 70, most were male (79%), white (87%), had primary bladder tumor (68%), and pure urothelial pathology (74%). Most received prior anti-PD1/L1 (92%) with pembrolizumab as the most common (72%); 56% received prior platinum-based chemotherapy with 77% cisplatin-based. Median prior therapy lines was 2. ORR to erda was 32% (12/37; all partial responses); another 12 pts had stable disease as best response. Median PFS was 7.5 months (mos) and median OS 18 mos. CDK2NA/B or MTAP loss was seen in 15 pts (38%) and ORR in these pts was 35%. 33 pts had canonical FGFR2/3 alts (S249C, Y373C, R248C, and FGFR3-TACC3 fusion) and 6 pts had non-canonical alts (A248C, A391E, Y375C, V555L/T807L, and JAKMIP1-FGFR3 fusion) with no significant difference in PFS (HR 0.8; CI 0.29, 2.3; p = 0.66), OS (HR 0.4; CI 0.06, 2.3; p = 0.086) or ORR (OR 2.1; CI 0.27, 27.7; p = > 0.99). Subgroup PFS/OS analysis in Table. Conclusions: Erda showed efficacy similar to clinical trial data. In this retrospective study with notably underpowered analyses, non-canonical FGFR2/3 alts or CDK2NA/B or MTAP loss did not seem to effect outcomes. Larger prospective studies, including granular NGS data, are needed to further validate “real world” data and further explore predictive biomarkers for erda to inform optimal therapy sequencing. [Table: see text]